Skip to content

Latest commit

 

History

History
217 lines (169 loc) · 8.03 KB

README.md

File metadata and controls

217 lines (169 loc) · 8.03 KB

bashful

This is beta quality! Use at your own risk.

Use a yaml file to stitch together commands and bash snippets and run them with a bit of style. Why? Because your bash script should be quiet and shy-like (...and not such a loud mouth).

Image

"But why would you make this monstrosity?" you ask... because &>/dev/null and | tee -a some.log and set -e; do something; set +e and other similar things is getting annoying. And besides... why shouldn't your bash script output look pretty?

Features:

  • Run bash snippits ('tasks') in series or parallel
  • A nice vertical progress bar to show current task status (with inline realtime stdout/stderr)
  • Download url references to bash scripts and executables
  • Bundle set of scripts and listed resources into a single simi-portable executable
  • Log all task stdout/stderr to a log file
  • ETA for running tasks (that have already been run before)
  • Configuration yaml block to control the behavior/look & feel
  • Detailed error reports when commands fail
  • Control which failing command should halt execution

Installation & Usage

go get github.com/wagoodman/bashful

There are a ton of examples in the example/ dir, but here are a few:

  1. The simplest of examples:
tasks:
    - cmd: echo "Hello, World!"

To run it:

bashful run hello.yaml
  1. A more realistic example: a build and deployment description
# ci.yaml
tasks:
    - name: Building app
      cmd: go build -ldflags '-linkmode external -extldflags -static -s'
      tag: build

    - name: Packaging app
      cmd: docker build -t my-awesome-app:v1 .
      tag: build

    - name: Publishing image
      cmd: docker push my-awesome-app:v1
      tag: deploy

    - name: Deploying app
      cmd: kubectl run my-awesome-app --image=docker.io/wagoodman/my-awesome-app:v1 --pt=80
      tag: deploy

Run all of the tasks...

bashful run ci.yaml

...Or run just the build steps:

bashful run ci.yaml --tags build
  1. Have an installer run things in parallel...
# install.yaml
tasks:
    - name: Installing bridgy
      parallel-tasks: 
        - cmd: sudo apt-get install -y tmux sshfs
        - cmd: pip install --user bridgy

...Or make an installer that downloads and runs everything it needs:

# install.yaml
tasks:
    # The given url gets executed by default
    - name: Installing Cuda and Bazel
      url: https://raw.githubusercontent.com/jasimpson/tensorflow-on-aws/master/toa_part_1of2.sh

    # Or you can control how the url asset (<exec>) gets used
    - name: Installing Tensorflow
      url: https://raw.githubusercontent.com/jasimpson/tensorflow-on-aws/master/toa_part_2of2.sh
      cmd: source ~/.bashrc && <exec>

Package up the installed into a single executable and give it to someone else to run:

bashful bundle install.yaml
# now you have a new executable called "runner", which can simply be executed
./runner

There are a ton of examples in the example/ dir. Go check them out!

Configuration Options

Here is an exhaustive list of all of the config options (in the config yaml block). These options are global options that apply to all tasks within the yaml file:

# this block is used to configure the look, feel, and behavior of all tasks
config:
    # which character used to delimintate the task list
    bullet-char: "-"

    # hide all subtasks after section completion
    collapse-on-completion: false

    # by default the screen is updated when an event occurs (when stdout from
    # a running process is read). This can be changed to only allow the 
    # screen to be updated on an interval (to accomodate slower devices).
    event-driven: false

    # This is the character/string that is replaced in the cmd section of a task to reference a downloaded url
    exec-replace-pattern: '<exec>'

    # the number of tasks that can run simultaneously
    max-parallel-commands: 4

    # log all task output and events to the given logfile
    log-path: path/to/file.log

    # show/hide the detailed summary of all task failures after completion
    show-failure-report: true

    # show/hide the last summary line (showing % complete, number of tasks ran, eta, etc)
    show-summary-footer: true

    # show/hide the number of tasks that have failed in the summary footer line
    show-summary-errors: false

    # show/hide the number of tasks completed thus far on the summary footer line
    show-summary-steps: true

    # show/hide the eta and runtime figures on the summary footer line
    show-summary-times: false

    # globally enable/disable showing the stdout/stderr of each task
    show-task-output: true

    # Show an eta for each task on the screen (being shown on every line with a command running)
    show-task-times: true

    # globally enable/disable haulting further execution when any one task fails
    stop-on-failure: true

    # This is the character/string that is replaced with items listed in the 'for-each' block
    replica-replace-pattern: '<replace>'

    # time in milliseconds to update each task on the screen (polling interval)
    update-interval: 250

The tasks block is an ordered list of processes to run. Each task has several options that can be configured:

tasks:
    - name: my awesome command      # a title for the task
      cmd: echo "woot"              # the command to be ran (required)
      
      collapse-on-completion: false # hide all defined 'parallel-tasks' after completion
      event-driven: true            # use a event driven or polling mechanism for displaying task stdout
      ignore-failure: false         # do not register any non-zero return code as a failure (this task will appear to never fail)
      show-output: true             # show task stdout to the screen
      stop-on-failure: true         # indicate if the application should continue if this cmd fails 
      
      parallel-tasks: ...           # a list of tasks that should be performed concurrently
      
      for-each: ...                 # a list of parameters used to duplicate this task
      
      url: http://github.com/somescript.sh # download this url and execute it
      md5: ae8abe98aeb389ae8b39e3434bbc    # an expected md5 checksum of the url provided

      tags: something               # one or more 'tags' that can be used to execute a sub-selection of tasks within a run yaml
      tags:                         # e.g. 'bashful run some.yaml --tags      something' 
        - something                 #      'bashful run some.yaml --tags      something,else'
        - else                      #      'bashful run some.yaml --only-tags something'

There are a ton of examples in the example/ dir. Go check them out!

Runtime Options

USAGE:
   bashful run [options] <path-to-yaml-file>
   bashful bundle <path-to-yaml-file>

COMMANDS:
     bundle   Bundle yaml and referenced url resources into a single executable
     run      Execute the given yaml

BUNDLE OPTIONS:
    None

RUN OPTIONS:
   --tags value       A comma delimited list of matching task tags. 
                      If a task's tag matches *or if it is not tagged* then it will be executed (also see --only-tags).
   --only-tags value  A comma delimited list of matching task tags. A task will only be executed if it has a matching tag.

GLOBAL OPTIONS:
   --help, -h     show help
   --version, -v  print the version

Wish list

All feature requests are welcome!

  • at least 70% test coverage
  • Multiple (serial) commands for a single task (cmd: [run something, run another thing])
  • Multiple url references for a single task (url: [https://someurl.com/some-script.sh, https://anotherurl.com/another-script.sh])
  • Allow parallel tasks within parallel tasks (really just allow deeper nesting of any kind of task)
  • Interact with the mouse to see more/less tasks (https://godoc.org/github.com/nsf/termbox-go#Event)